View
3
Download
0
Category
Preview:
Citation preview
Big data, digitalisation and machine
learning have all become buzz words in
their own right in the past five or so years.
Their presence in the mining zeitgeist has led
to the largest companies installing nodes of
various types on everything from haul trucks
and drill bits to ventilation fans and flotation
cells. The data generated from such equipment
is certainly ‘big’, but whether it is all of
economic value is less clear.
“The idea that data is intrinsically valuable is,
in my opinion, not true,” Penny Stewart, CEO of
Australia-based PPetra Data Science, told IM.
“Some data hasn’t got value as it is not
relevant to a mining company’s core business
drivers.”
Mining companies, arguably, need to treat big
data investments like project development
decisions where, if a proposed mine provides
the necessary payback, the investment is
greenlit. Any proposed project investment that
doesn’t reach the required threshold,
conversely, is put on hold.
This type of investment rationale would
ensure mining companies get the required
returns on their big data packages and only
spend money where they need to.
Stewart is not the only one that believes big
data requires more than just the data itself to
make a strong value proposition.
Andrew Jessett, CEO of monitoring and
automation solutions provider MMineWare (part
of KKomatsu Mining), told IM: “Adopting new
technologies to increase the use of big data is
one thing…[But] to deliver real, sustainable
performance improvement, the digitisation of
mining must lead to joined-up process and
equipment improvement across the whole value
chain.”
“Smart digitisation”, as Jessett says, “means
sharing data across the mining value chain and
turning it into actionable, accessible information
to improve operational performance”.
With many operations suffering from ‘silo
syndrome’ and the interoperability issues that
come with this, providing an integrated
operation where data and analysis flows freely
is not as easy as one may expect.
Profit from the plantStewart says: “There is two parts to how you
apply big data in mining, in my mind. One is
around the process and getting maximum
production and efficiency out of a process. In
62 International Mining | NOVEMBER 2018
Leverage the learning
As the name suggests, big data is a vast and still-expanding sector that has been catching on in mining ascompanies look to squeeze more value out of existingassets. Dan Gleeson gets to the bottom of how thosesupplying these solutions are impacting the bottom line
BIG DATA AND DIGITALISATION
Synertrex provides fact-based insights thatallow operators to
optimise equipmentperformance and
increase throughput,according to Weir
Mineware is working with partners such as Komatsu Mining, Modular Mining and RPMGlobal toenable and deliver full integration between the companies’ systems
that case, capturing all of the data from all of
the equipment involved in the process efficiency
and productivity is relevant.
“But, if you are talking about asset
maintenance and how much value there is in
applying big data analysis on non-critical
assets, it might not be as important in mining,
where you have a lot of backup equipment in
existing circuits.”
Where there is a case to be made for the
former is the process plant, according to Stewart.
“This is an area where you can get a lot more
value from integrating data and modelling a
whole process,” she said.
The type of data generated out of the
processing plant – time-series data – is also
suitable for the type of machine-learning
algorithms starting to be applied in the mining
industry.
“The other advantage with the process plant
is you can deploy machine-learning algorithms
using existing servers that are already
supporting the process plant. You…[also] don’t
have to put everything in the cloud if you don’t
want to,” Stewart said.
Petra recently collaborated with mining
company PanAust to help predict tailings grade
at the Ban Houayazai gold-silver mine in northern
Laos using its MAXTA digital twin technology.
The machine-learning algorithm provided to
PanAust integrated two years of 3D geological
data with plant data to derive a formula to apply
to block models, allowing both ‘backward’
reconciliation analysis and ‘forward’ predictions.
While similar digital twins could be
developed without the geological information at
the start of the process, Stewart admitted this
would reduce the accuracy of the results. So,
again, integration is key in this example (you can
read more about how Rio Tinto plans to use a
digital twin at its Koodaideri project on page 60).
OEMs are also catching onto these new big
data analysis solutions in the process plant as
they look to increase the availability of their
products.
One such company is WWeir, which officially
launched its Synertrex® industrial internet of
things (IIoT) platform earlier this year to
improve the uptime and performance of its
portfolio of mining equipment.
Synertrex® provides fact-based insights that
allow operators to optimise equipment
performance and increase throughput,
according to the company.
The platform has been developed by Weir
with the help of leaders in IIoT, data science and
edge computing. This includes the likes of
Microsoft, Dell and Petra Data Science.
“Weir has developed the Synertrex® platform
as a highly specific augmentation of their key
products,” the company said.
Alasdair Monk, Group IoT Product
Management Director for Weir, told IM: “The
development of our Synertrex® platform is an
evolution of Weir’s equipment and services. It
allows our customers to monitor their
equipment in real time, enhancing the
performance and reliability of their operations.
“Weir offers a series of service packages as
standard. With the Synertrex® platform, we have
redefined our service offering, advancing our
reach, responsiveness and understanding to
customer’s critical issues and opportunities.”
At the moment, monitoring of Warman®
pumps, Cavex® hydrocyclones, GEHO® pumps
and Enduron® HPGRs, screens and crushers can
be enhanced with the technology.
While it is still early days for Synertrex®, a
number of success stories have already emerged
from several installations across the globe.
This includes ascertaining the causes and
advising a mine site in Mexico on the required
corrective actions for two GEHO® hydraulically-
driven positive displacement pumps – all
without sending a service technician to site.
At a mine site in South Africa, the Synertrex®
platform detected a change in the operating
performance of an Enduron® screen. This prompted
a detailed inspection which found hidden cracks on
a side panel and suspension springs, which was
quickly addressed. On the same site, the Synertrex®
platform detected an issue with an adjacent crusher
which fed the screen.
BIG DATA AND DIGITALISATION
Keep on track
United. Inspired.
Utilize machine data to boost productivityProactively monitor your machines to continually assess performance
and address potential problems. With a variety of software and hardware
monitoring options available, data is accessible whenever you need it to
analyze and optimize performance. epiroc.com
“This demonstrates how Synertrex®
technology can provide an advanced
understanding of the entire process plant and
identify anomalies impacting multiple pieces of
equipment,” the company told IM.
According to Monk, this hat-trick of
discoveries by the Synertex® platform saved the
customer “millions of dollars in lost production”.
Metso is also venturing into this space with
the development of its own advanced analytics
and AI-backed platform for comminution circuits
(read more about Metso Metrics for Mining
below).
More value from the fleet There are plenty of other applications outside of
the plant where analysis and interpretation of big
data can be used to create value for mining
companies.
Fleet management and maintenance is one
area where numerous companies offer solutions.
The key to these is integration between
mining operation systems, according to
Mineware’s Jessett.
“Traditionally, data has been very fragmented
in the operations space. To reap the benefits of
big data, mining companies and vendors need to
break down these individual operating siloes,”
he said.
Jessett and Mineware are not just paying lip
service to this; they are working with partners
such as Komatsu Mining, MModular Mining
Systems and RRPMGlobal to enable and deliver
full integration between such systems.
“Our comprehensive integration between
RPM’s short-term scheduling solution, Xecute,
and our shovel and excavator monitoring system,
Argus, is a world-class example,” Jessett said.
This combines office-based planning and
scheduling systems with in-pit loader technology
to enable mine planners and equipment
operators to compare planned production
activities against what is happening on site in
real-time.
“Together, Xecute and Argus create a living
mine plan that continually evolves and closes the
feedback loop between mine planning,
scheduling and operations to maximise
productivity through greater control, flexibility
and predictability,” he said.
Wenco Mine Systems’ Business Intelligence
Product Manager, Simran Walia thinks fleet
management and maintenance data need to be
factored into the same analysis package for
complete solutions.
“Productivity per shift, loads per shift,
equipment condition, payload analysis – they all
involve several operational systems and they’re all
interrelated, so analytics tools need to show that
relationship, the cause and effect,” he told IM.
Wenco thinks it has solved this with its Avoca
Mining BI platform, which, according to Walia,
takes all parts of the “mining process into
consideration, so processing and mining teams,
maintenance and geologists all get the value
from big data”.
Avoca digests this data and produces
visualisations of it to highlight trends within the
mining process that teams can act on.
Dingo believes integration of a different type is
required to ensure these big data tools provide
the value mining companies require.
“At Dingo, we know that technology alone is
not the answer,” the company told IM. “It’s
integrating technology into the way people work
and keeping the people who use the technology
central to the process.”
By sticking to these philosophies, Dingo has
provided its customers, on average, a more than
4:1 return on investment in parts costs alone
under its fleet maintenance solutions, according
to the company.
“One example is a small surface mine in
Nevada which had a target to reduce their
maintenance budget by 15%. By partnering with
Dingo, they achieved a 24% reduction in year
one of their programme,” Dingo said.
The company’s Trakka® predictive
maintenance software helps manage and
correlate multiple condition monitoring
technologies into one database to aid analysis
and decision making.
On top of this, Dingo has a team of
64 International Mining | NOVEMBER 2018
BIG DATA AND DIGITALISATION
The Umbrella of Mine Safety
www.dsiunderground.com
Delivering the Support You Need
maintenance experts, working under its
Condition Intelligence® platform, who use Trakka
to help filter through the massive amounts of
data coming through from monitoring equipment.
This type of service is extremely valuable; in
effect offering a human touch to the machine-
heavy processing of data.
And, the company is offering “actionable
information on the go” with the development of
products such as the Trakka Field Inspection App
and Asset Heath Management Dashboard and App.
The increased uptake and use of such apps
has led to higher volumes of digital reports,
images and other files; something the company
addressed in the release of Trakka 4.5.12. This
features drag and drop functionality, flexible CSV
data import, a new interface to dozer tracker
inspections and new ERP integration web service
– all geared towards simplifying the user experience.
Simon Van Wegen, Product Manager,
Maintenance Systems, Modular Mining
Systems, says there are four main areas
maintenance departments must focus on when it
comes to the health of equipment components.
These are:
� The failing component;
� The root cause of the component’s failure;
� Component health degradation for the failure
type (is the failure detectable earlier on the P-F
curve? (relationship between a potential
failure and a functional failure)), and;
� Actions to take to prevent future failures, or a
plan to replace the component before it fails.
“An intelligent maintenance management
system provides plenty of data to confidently
identify the above and some systems can often
process this data by two different technological
means: edge computing and cloud computing,”
Van Wegen told IM.
“The MineCare Maintenance Management
system from Modular Mining Systems, for
example, utilises both of these computing means
to help increase equipment availability, minimise
troubleshooting efforts, mitigate catastrophic
failures, and provide analyses to identify
component health degradation and predict when
potential failure points will occur.”
The MineCare system leverages edge
computing’s real-time capabilities to quickly
notify operators and maintenance teams about
critical, time-sensitive conditions while also
enabling rapid data transmission for central
processing, according to Van Wegen.
“For example, if a haul truck’s engine oil
levels are critically low, the MineCare system
will immediately notify a maintenance
technician. This real-time information drives
technician action to stop the unit and address
lubrication levels before major engine damage
occurs,” he said.
For large quantities of real-time and batch
data that could have any influence over a
NOVEMBER 2018 | International Mining 65
BIG DATA AND DIGITALISATION
The release of Dingo’s Trakka 4.5.12 comes withimproved drag and drop functionality, flexibleCSV data import, a new interface for dozertracker inspections and new ERP integrationweb service
WWW.D INGO.COM
The Global Leader in Predictive Maintenance
66 International Mining | NOVEMBER 2018
component’s health, this is where cloud
computing comes in.
“Cloud computing’s massive storage capability
also facilitates easier machine learning and
analytics. Gone are the days that require an
engineer to sift through huge raw data files; the
MineCare solution feeds large datasets into
intensive machine-learning algorithms that can
identify trends, calculate predictions about
components failures, identify components that are
operating differently than other components with
similar operating hours, and much more,” he said.
This is a potent combination that has brought
about some impressive results.
For example, a large mine in Australia reduced
its unplanned events by about 25% in roughly
one year after leveraging the MineCare system’s
intelligent algorithms to identify potentially-
problematic trends before they resulted in
component failure.
And, on top of this, Modular is offering a
remote monitoring package, RemoteCare, where
highly-trained specialists monitor and manage
real-time telemetry data generated by a mine’s
entire fleet, providing valuable insight that can
lead to valuable results.
“By collecting the right data at the right time
and analysing it at the appropriate edge or cloud
level, the MineCare system helps maintenance
teams resolve the root cause of repeated failures
to increase equipment uptime and decrease
associated maintenance costs,” Van Wegen
concludes.
Teck Resources is using such big data analysis
at its steelmaking coal operations in British
Columbia, Canada, incorporating artificial
intelligence into its data generation procedures.
“Through our partnership with GGoogle Cloud
and PPythian, we are unlocking new insights from
the millions of data points generated by our
mobile fleets,” the company said.
“Issues that were previously unpredictable,
BIG DATA AND DIGITALISATION
This month sees the launch of Metso Metrics for Mining, a cloud-
based IoT solution designed to monitor and predict integral
components of a mine’s comminution circuit. Dan Gleeson received
some more details on the new innovation from Metso Chief Digital Officer
Jani Puroranta ahead of its official debut.
IM: How will Metso Metrics for Mining differ to Metso Metrics for
Aggregates?
Jani Puroranta: Metso Metrics for Aggregates is more akin to other fleet
management solutions you can find for various kinds of mobile equipment.
It uses satellite connection (or cellular internet, when available) to
communicate with Metso’s backend. Hence, it collects only small amounts
of the most critical data such as operating hours, alarms, etc.
Metso Metrics for Mining, on the other hand, collects vast amounts of
data from circa-100 sensors within the machines at one second resolution
or even higher. Connectivity is typically fixed internet, which allows for
much more data to be transferred to Metso’s cloud. The aim is to monitor
the current condition of the machine and its various sub-systems
(hydraulics, lubrication, etc), and to predict critical component failures
already in advance. With the help of advanced analytics and AI, we can
automate some of these monitoring and prediction tasks.
Also, the data allows Metso experts to look for ways to improve the
machine’s operation such as how to reduce the specific energy
consumption, or to find correct operating parameters for the application to
maximise throughput and product quality.
Hence, in summary, Metso Metrics for Mining is a more technologically
advanced solution combined with a ‘hypercare’ expert support model.
Metso Metrics for Aggregates is more a self-help tool for the user.
IM: You have mentioned one of the major roadblocks for predictive
maintenance and AI in fixed plant applications in mining has been the
non-standardised nature of these plants. How is Metso getting around this
problem?
JP: Unfortunately, there is no easy way out of this since ultimately we need
to work with heterogeneous fleets of all ages. Hence, prioritisation of
development work has been key for us to move forward at good speed. We
started the work with Metso’s primary gyratory and cone crushers, as well
as vibrating screens. Next in line will be other asset types such as mills
and pumps. We always try to see where we can add most value to the
customer’s operations and where there is little or no relevant offering on
the market today. What we do not want to do is repeat something that
already exists (such as APC or GMD condition monitoring), but find blank
areas within the processing plant. These are not hard to find – the tough
part is how to solve them remotely.
IM: Can you give any more insight into the African, US, South American
and Australian operations currently trialling Metso Metrics? How has it
impacted the uptime of the crushers, for example?
JP: We have been piloting Metso Metrics on five customer plants during
the past year. On these sites, the biggest benefits have actually come from
collaboration between Metso’s and the customer’s experts. In the past,
both were operating blindly without really knowing what is causing
maintenance issues, critical component failures or reduced throughput.
Now, we have the data to get to the root causes quickly so that machine
and circuit performance can be quickly improved.
However, it is still a bit too early to show statistically reliable numbers.
And sometimes you cannot really quantify the benefit of a predicted failure
that was avoided with preventive maintenance – how can you measure
something that never happened?! But with the connected fleet expanding
to dozens of machines of the same type, we should be able also to put
numbers on the benefits over the coming 12-24 months.
The P-F curveidentifies therelationshipbetween apotential failureand functionalfailure, based onequipmentcondition over time
such as potential electric failures, are now being
identified before they happen by machine-
learning algorithms. We are also modelling and
predicting remaining life span of our trucks,
determining wear, identifying abnormal failures
and enhancing alarm and notification systems.”
This type of machine learning is helping
minimise unplanned maintenance, reduce overall
maintenance costs and extend equipment life,
according to Teck
“It is estimated that at one site alone there is
potential for over C$1 million ($769,843) in
annual savings from implementing this
programme,” the company said.
InteroperabilityWhile there is no shortage of potential data
points at most mines for machine-learning
algorithms to study, there are still barriers
holding back data integration in this space – the
main one being interoperability.
The problem boils down to many of the
machines involved in the mining processes
speaking a different ‘language’.
One solution to this problem is to come up with an
industry standard that all machines use, which
facilitates machine-to-machine interaction and allows
for smooth transfer of data between mixed fleets.
IREDES and the Global Mining Guidelines
Group are two organisations pushing for such a
standard.
Chairman of the IREDES initiative, Christoph
Mueller told IM: “The reasons why
standardisation is a hot topic now…is the fact
that the operating mining companies realise if
they join up with one manufacturer for their
autonomous fleet, they are locked to this specific
OEM, and there is no chance to include other
equipment, even if ‘their’ OEM does not supply
this kind of special equipment.”
Mueller is aware there is a conflict between
opening up data exchanges in this way – causing
potential intellectual property issues – but says,
for process information to be properly and
quickly used, an industry standard data
communication interface is required.
“Today, information processing capability is so
big that, in my opinion, it makes much more
sense to have the raw data turned into really
usable information right on the machine and use
this information for maintenance, operations and
quality assurance.
“This way, the machine by itself can compute
information like estimated time to next service,
load integrals for all kinds of different
components, etc in order to allow cost-effective
maintenance and to optimise operational
dispatch availability.”
Mueller said the initiative has seen more
interest in implementing the IREDES profiles on
machinery in the last two years and it was in the
process of carrying out a review of the standard –
news of which is likely to be discussed at the
upcoming APCOM conference in June.
Even without a standard machine-to-machine
language, OEMs and data analysis providers are
still integrating their own software and hardware
with other system providers’ systems.
Technology company RRCT has found a way
around proprietary systems to install its
automation and teleremote solutions at many
mines around the world.
Brendon Cullen, RCT’s Automation & Control
Product Manager, said the company likes to
work with OEMs and dealerships to ensure they
are comfortable with the programmes they
provide mining customers, but when information
is not forthcoming, they can go down another
route.
“RCT has a skilled team that are able to
reverse engineer a solution by interrogating the
machine’s data sources,” he told IM.
“Interoperability is key for RCT as our ability to
fit similar systems to many makes and models of
machinery on a site and integrate with their
current fleet management, dispatch and
information systems, is what sets us apart from
our competition.”
Cullen points out there is no OEM machine
RCT has not been able to integrate with over the
years – from a 1980’s dozer to a 2018-built
underground truck.
And, more OEMs are opening up data sources
BIG DATA AND DIGITALISATION
Building tomorrow’s technologiesfor today’s mines
www.modularmining.comMAINTENANCE MANAGEMENT
PUT THE BRAKES
ON UNPLANNED MAINTENANCE
Increase reliability & MTBF
Maximize availability
Reduce maintenance costs
to companies like RCT, according to Cullen.
Caterpillar is one of these OEMs, as was
recently made clear on a project in the Pilbara of
Western Australia. Here, the company is on course
to complete a retrofit of Cat MineStar™ Command
for hauling autonomous technology on 24
Komatsu 980E haul trucks by the end of the year.
Bill Dears, MineStar Solutions Commercial
Manager, lays out a number of reasons why
today’s mining companies are demanding
interoperability and, in turn, why Caterpillar is
providing solutions to them:
� They have made significant investments in the
machines they already own and are not
interested in replacing whole fleets in order to
leverage new technologies;
� They rely on a variety of manufacturers and
vendors to provide their information
technology/operational technology systems,
production equipment and other support
vehicles and need technologies that can work
across all their machines, systems and
processes, and;
� They want to optimise their operations and
make them as lean as possible. The
connectivity that pulls together all machines
and systems, across the site and the
enterprise, is what helps mines achieve that
optimisation.
Petra’s Stewart thinks APIs (application
programming interfaces) and SDKs (software
developer kits) can provide another route around
the data language issue.
“Often when we integrate our MAXTA digital
twin solution, for example, we integrate other
companies’ data into our product (through an
API/SDK), but that one-off integration cost is
relatively small when spread across machine-
learning models of throughput, recovery and
product quality,” she said.
“The good thing about an API management
architecture approach is that it enables
companies to solve interoperability issues in
months, and not years.”
OEM-agnostic automation solutions provider
Autonomous Solutions Inc, too, is making the
most of API functionality in its bid to integrate
with, mostly, proprietary interfaces.
Drew Larsen, Director of Business
Development for ASI Mining, told IM: “While we
recognise ongoing efforts and initiatives around
interoperability are having some effect, we
anticipate universal standards and open
interfaces will be slow coming from OEMs and
other providers.
“ASI works around these proprietary interfaces
through various means, but generally seeks
cooperation from system providers through
supported APIs or access to protocols where
available. It also provides interface access to its
own systems to partners and project participants
to promote interoperability within an
autonomous environment.”
Whether it is an industry standard, an API, an
SDK, or something else, there is a clear need for
those providing data to the mining industry to
enable more integration and create the platform
for mining companies to leverage an ever-
increasing number of solutions.
68 International Mining | NOVEMBER 2018
BIG DATA AND DIGITALISATION
IM
Resolute Mining, in partnership with SSandvik, is in the throes of an
ambitious project to bring a fully autonomous underground gold mine into
production at its Syama Underground asset in Mali. Dan Gleeson spoke
with Resolute Chief Operating Officer Peter Beilby about why the company
decided to go down this innovative route.
IM: What makes Syama Underground suitable for automation?
Peter Beilby: It is a large orebody which will be operated as a sub-level
cave. Sub-level caves, by their very nature, lend themselves to automation
given the repetitive mining process. You continually go through a cycle of
drilling, charging and firing the blastholes, then loaders extract the ore
from drawpoints and, in our case, load into an ore pass.
These machines operate by laser scanning and learning the route they
need to take. Once they have done that initial survey, they can repeat the
route at speed – potentially at much higher overall speeds over time than a
manual operation.
IM: How much of a role does big data and digitalisation play in Resolute’s
plans for an automated mine at Syama Underground?
PB: Big data and digitalisation is a really important part of operating a
fully-automated mine. What is particularly exciting is that we’re just at the
beginning of understanding how powerful these tools will be. Our
automation team has been working with our partner, Sandvik, to help
further develop what they already have in place in terms of data
management and analytics for Syama. We will be generating a significant
amount of data from the machinery underground and it is how we actually
mine that data to come up with the information that will help us optimise
performance.
IM: What facilities have you got in place to manage data at the Syama
Underground project?
PB: We are in a very remote place, which is one of the challenges. We rely
on bandwidth associated with satellite connections.
We know we are going to need to transmit a large amount of data once
in operation – it is not just the data we will be collecting from the
machines, it is also the cameras we will have in place, ventilation controls,
ground controls, etc. We decided to proof ourselves for the next five to 10
years by using 120 pair fibre-optic cables down each decline to ensure we
don’t lose functionality and have enough capacity for future operations.
One of the interesting aspects of our operation is that by linking in to
Sandvik’s OptiMine solution you can...instead of, say, looking at five LHDs
on your mine site, compare your machines with Sandvik loaders all over
the world. That means you are really starting to hone in on exactly what
drives performance and identify areas for optimisation and cost savings
within your maintenance programme.
IM: Aside from the AutoMine and OptiMine systems, what other
information and production planning systems is the company using to
help automate operations?
PB: If we talk about the mine itself, we use DDeswik for mine planning. The
idea is that the database all feeds into this.
The hardware that is being put into the underground mine has the
ability to carry significant amounts of data. As such, fans, pump stations,
rockbreakers, geotechnical and other data-hungry usage such as video
underground will also be monitored and controlled through the same
system.
IM: Have you chosen to have a complete package from Sandvik in order to
alleviate any potential interoperability problems?
PB: When we started our Syama Underground journey two years ago, we
went out like every miner and said: ‘We want a yellow truck, an orange
jumbo and a different coloured loader’. This was based on our own biases
and preferences, and we envisaged a third party running them.
But, through that process, we looked at various partially-automated
mines around the world (Finch and Northparkes, included). This ignited
the fire in us to pursue automation further.
At that particular time, the only OEM that could automate the whole
process was Sandvik. We realised we needed to align ourselves with
Sandvik and leverage that relationship to deliver a fully automated
solution.
IM: What’s the latest timing on Syama?
PB: We are firmly on track to deliver the world’s first customised, fully
automated underground gold mine. Sub-level caving will commence in
December 2018 and we will then ramp-up to full automation and run-rate
production by June 2019.
Recommended